项目名称: 基于STDP机制的脉冲神经网络自组织功能研究
项目编号: No.61305077
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 自动化技术、计算机技术
项目作者: 周茜
作者单位: 河北工业大学
项目金额: 22万元
中文摘要: 最近几年来,人工神经网络的研究重点逐渐转向更具生物真实性的脉冲神经网络。脉冲神经网络采用脉冲编码(spike coded)而非频率编码,因此这种新型的神经网络可以获得更多的信息和具备更强的计算能力,而且在分布式计算更好的模拟了生物神经网络的处理方式。STDP突触可塑性机制是一种时间窗口非对称形式的Hebbian 学习,使Hebb学习算法更为精确,被认为与学习、记忆的神经网络机制密切相关。本课题研究基于STDP机制的脉冲神经网络,旨在提取不同结构脉冲神经网络中STDP机制对网络的调控规律,包括STDP机制对网络结构的动态调节,对神经元之间的协同作用的调节,为构造新型的自组织神经网络以及脉冲神经网络的具体应用打下基础;利用STDP自适应调节机制,构造输入-输出关系不随内部拓扑结构改变而改变的脉冲神经网络,为实现自组织功能电路做准备。
中文关键词: 脉冲神经网络;STDP机制;自组织;调控规律;
英文摘要: Recently spiking neural networks attract more and more research interest in artificial neural network field. These kind of neural networks encode information in the timing of single spikes, and not only just in their average firing frequency. Hence, they can encode more information. Spiking neural networks are more computational powerful and biologically more realistic in parallel processing pattern. Synaptic modification rule of STDP (Spike-timing-dependent plasticity) makes Heb learning more accurate and is known to be closely related to learning and memory process of brain function. In this project, how STDP dynamically regulates spiking neural network structure and the cooperative activity between neurons will be studied. It will be useful to the design of better adaptive neural network and the application of spiking neural network. Moreover, the project will try to determine spiking neural network with specific function, that is the input-output relation of the network will not alter with the change of internal structure. It will make preparations for the implementation of adaptive circuit.
英文关键词: spiking neural network;spike-timing-dependent plasticity;self-organization;regulation rule;